Professional Documents
Culture Documents
*
Zhongxiu Zhao and Kevin Honglin Zhang
Abstract
*
Zhongxiu Zhao: School of International Trade and Economics, University of International Business and
Economics, Beijing, 100029,China.
Kevin Honglin Zhang: Department of Economics, Illinois State University, Normal, IL 61790-4200, USA.
We wish to thank Yufei Liu, Zheng Xu, Husniddin Mustafaev, and Wayne Jiao for valuable assistance in data
entry. Of course, the usual caveat applies.
1
1. Introduction
In the recent literature on international economics and economic growth, the link between
technology transfers and foreign direct investment (FDI) made by multinational corporations
(MNCs) seems to have been prominent. Theoretically, there is a widely shared view that
technology may be transferred to host developing economies through (a) MNCs’ backward and
forward linkages with indigenous firms and customers; (b) imitation of domestic firms by
“learning by watching” in the presence of MNCs; (c) induction of trained workers and managers
by MNCs; and (d) relocation of MNCs’ R&D activities to host economies. On the other hand,
however, it is sometimes suggested that MNCs may (a) restrict diffusion of technology
(especially advanced ones) to their subsidiaries abroad; (b) transfer technologies that are
inappropriate for the host country’s factor proportions; (c) prefer imports of key
components/parts from parent factories to local suppliers, reducing linkage effects; and (d)
maintain their technological advantage by forcing host economies to follow strict rules of
intellectual property rights. Although further theoretical insights would be valuable, empirical
analyses of the issue are needed as well for a better understanding of relationship between FDI
While cross-country studies are useful, single-country research, despite its shortcomings,
yields a better indication of the detailed picture for each case and provides practical policy
implications. In this context, it is useful to note that three recent single-country studies, which
used industry- or firm-level data for small developing economies or developed countries for the
1970s and 1980s, reached fairly divergent conclusions (Aitken and Harrison, 1999; Blomstrom
and Sjoholm, 1999; and Potterie and Lichtenberg, 2001). No study has been conducted for a
2
large developing country with substantial FDI inflows for the 1990s. Thus there is a need for
This article seeks to fill this gap by using the Chinese industrial data for the 1990s. The
case of China is important empirically because of its huge magnitude of FDI inflows and the
issue being largely unexplored.1 Besides the use of a large sample of 187 industries and the focus
on the 1990s, this study has several distinctive features. First, it specifies empirically two
channels through which FDI might transfer technology to local industries: direct effects are
participation in the sector to which the industry belongs. Second, it studies the host country’s
absorptive capability to capture benefits from FDI, which is believed to be one of the key factors
includes most major influences on technology transfers from FDI. Last, in addition to the entire
sample, it considers estimates for the subgroup of labor- and capital-intensive industries to detect
2. The Model
As noted in Section 1, the nexus between FDI and technology transfers has been explored
empirically by several scholars. For instance, B. Aitken and A. Harrison (1999) used firm-level
data for Venezuela over the period 1976-89 to study if technology “spillovers” existed from
foreign to domestic firms in case of different foreign equity participation. They concluded that
technology “spillovers” seem to be small from foreign to small local enterprises or even negative
1
The cumulative FDI into China over the period 1979-2005 is $620 billion, with over $60 billion inflows annually
in 2004 and 2005 (SSB, 2006). China has been the largest FDI host country in the developing world since the early
1990s, with a share of over one third. For a few years in the early 2000s, China even surpassed the US and received
the largest amount FDI inflows in the world (UNCTAD, 2005). While there have been some discussions on
technology transfers from multinational corporations in China (for example, UNCATD, 2002), they are limited to
qualitative descriptions. Few studies have been devoted to quantitative analyses of the issue.
3
from joint ventures to local firms. Based on Indonesian micro data of 1991, M. Blomstrom and F.
Sjoholm (1999) found that domestic establishments benefit from technology spillovers of foreign
establishments’ high productivity, and that foreign ownership does not affect the spillovers. B.
Potterie and F. Lichtenberg (2001) used the country-level data of inward and outward FDI for
US, Japan and EU over the period 1971-90 and indicated that inward FDI does not transfer
technology to host economies and outward FDI transfers technology only if a country invests in
The models used in the aforesaid studies are broadly similar, and we also employ a
parsimonious specification that resembles the ones adopted by these researchers. We start with a
added and intermediate goods, and technical changes are value-added-augmenting. Therefore a
value-added production function for an industry may take the following form:
where Qi denotes the value-added in industry i, Li and Ki are the input of labor and capital, Ai
represents technical factors associated with FDI, and β1 and β2 are the elasticities of labor and
Specifically, the technical factor term (Ai) can be decomposed into three technological
components as follows: 2
2
The term technology used here refers to the knowledge that is embodied in products, processes and practices.
Products comprise the knowledge of how things work, their design, and their interface with other products.
Processes comprise the knowledge on how a product can be produced or changed. Practices consist of the routines
necessary to manage the product-process combination and the knowledge regeneration process.
4
where Fi is the technical factor associated with direct effects of technology transfers from FDI in
industry i, Si is the technical factor associated with spillovers of FDI in the industrial sector, and
Ri is the technical factor associated with FDI absorptive capacity in the industry. Being the
elasticities of F, S, and R, respectively, β3 thus measures the direct impact of FDI on the
technology of industry i, β4 captures externalities of FDI in the sector on the technology of the
industry, and β5 indicates the technological catch-up ability in the industry and its
complementarities with FDI. Replacing Ai in Eq. (1) with Eq. (2), we get the production function
Addition of a constant term and a stochastic component to Eq. (3) yields the following
ln Qi = β 0 + β1 ln Li + β 2 ln K i + β 3 ln Fi + β 4 ln Si + β 5 ln Ri + ε i (4)
Eq. (4) constitutes the basis for our econometric analysis of the FDI-technology transfer
link with the Chinese industrial data. The rationale of the FDI-technology transfer link for the
specification is based on the direct and indirect impact of MNCs on the technological
development in host industries (Caves, 1996; Das; 1987; Markusen and Venables, 1999; and
Teece, 1977). The direct transfers take place by introducing capital goods (equipment), new
processing practices, new products, and new management skills. The indirect transfers are what
MNCs may not intend to do, or spillovers. Such spillovers include backward and forward
While the potential for MNCs to transfer technology exists, the benefits to host countries
do not automatically accrue. The magnitude and extent of technology transfers may be related to
host industry characteristics. Especially the level of absorptive capability is needed to acquire
5
and work with the technology. For example, technologies from MNCs may not be appropriate
for local firms and industry and thus may not enable them to compete effectively in the global
market. Local firms and industries have to make a variety of investment to actually benefit from
technology inflows. Therefore the host capability to absorb foreign technology turns out to be an
It is possible for host technological capability even to be worse off with inward FDI. The
technology provided by MNCs may have little impact on domestic technological development
and may in fact inhibit their development by stifling the growth of indigenous entrepreneurship
as a result of the MNCs dominance of local markets. Thus local development in technology may
be suppressed by MNCs and “crowding-out” effects may occur. Moreover, MNCs may not
intend to transfer technology to host countries because they wish to maintain their status of
technological monopoly.
All data employed in the work are taken directly from The Third National Industrial
Census of People’s Republic of China (SSB, 1997). This is the most recent industrial survey that
followed the international standard classification and covered all industries in China. The
industrial survey was conducted by China’s State Statistic Bureau in 1995, and the data collected
from the survey were published in 1997. The 191 industries are categorized into 37 sectors.3 The
dependent variable (Q) is measured by current value added of an industry, and L is taken as total
number of employees in the industry. Domestic capital stock (K) is taken as the current value of
total domestic capital formation in the industry. Effects of direct technology transfers from FDI
3
The 37 sectors and 191 industries are listed in Appendix.
6
(F) are measured by percentage of current FDI stock in total capital stock. Spillovers of FDI (S)
are proxied by the average share of foreign equity participation in the industrial sector to which
the industry belongs, weighted by the industry’s output. The absorptive capacity (R) is measured
In addition to the estimation of Eq. (4) for the full sample, we run regressions for two
technology transfers from FDI may be different due to factor intensity in industries. The criteria
used for the division is the capital-labor ratio of ¥23,000 (RMB of the Chinese currency) per
worker, which is consistent with the Chinese government’s classifications of industries as well as
international standards.4
For the purpose of comparison, we run an additional regression in each case, which
includes three FDI indices as the sole explanatory variables to show to what extent FDI may
Before regression estimates of Eq. (4) are presented, it is appropriate to provide some
descriptive statistics for the sample on which the regressions are based. Although the total
number of industries covered in the study is 191, missing data reduce the sample size slightly to
178. Every industry for which data for the relevant variables are available in the source cited has
been included. Thus, there is no direct selection bias in the sample. Table 1 contains descriptive
statistics for all of the variables corresponding to 178 observations of the full sample and the two
4
For examples, the capital-intensive industries include crude oil extraction, crude oil processing, organic chemicals,
chemical pharmaceutical preparation, chemical pharmaceutical products, biological products, synthetic fibers,
automobiles, aerospace, electrical industrial apparatus, communication equipments, electronic computers, electronic
apparatus, and office instrument machinery. The labor-intensive industries include coal extraction and washing,
canned food, cooking sauces, cotton textiles, woolen and silk textiles, clothes, hats, footwear, wood furniture, paper
processing, paper products, toys, and game equipment.
7
The Appendix lists the sample industries by 37 sectors and identifies the capital-intensive
Table 2 presents the main regression results for the full sample and the two sub-samples.
For each case, estimates are also reported for the regression with the three FDI indicators as the
(1) The explanatory power of these models is quite high for the diverse cross-industry
sample. All of the three models (with the full sample, the capital-intensive sample, and
the labor-intensive sample) explain nearly all variability across industries, as indicated by
the high level of R2 (0.98). The fit of the regressions is good with F-statistics being
(2) The three FDI variables seem to be important in explaining China’s industrial
performance. The explanatory power is high (R2 = 0.92) for the three regressions that
include only the FDI variables, suggesting that over 90% of the variance in the value-
(3) The evidence in favor of direct technology transfers from MNCs is lacking, instead, we
found the opposite impact. In all estimates, the term for direct FDI effects (F) has the
“wrong” (negative) sign with significance at the 1% level. The estimates suggest that the
large (small) share of foreign investment in an industry is associated with low (high)
productivity of the industry. One possible explanation is that inward FDI yields a net
negative effects on the Chinese industries due to the strong MNC market power and
“crowding-out” effects. The other explanation is that the positive FDI-technology transfer
8
link found in some studies may reflect the tendency that MNCs generally locate and
invest in more productive industries (Aitken and Harrison, 1999). In fact, the industrial
composition of inward FDI in China is different from the pattern in many host economies.
products were mainly for exports (Zhang, 2000; 2001a; 2001b; and 2004).5 Therefore,
the negative effects from the estimation may not suggest a decline in domestic
productivity with the foreign participation in industries, but the correlation of FDI with
(4) The contribution of MNCs to technology transfers in China seems to be mainly through
indirect effects (spillovers). In all cases, the parameter for the FDI spillovers term (S) is
robustly positive, as many scholars have reported (for example, Blomstrom and Sjoholm,
1999). The positive effects are large and statistically significant at the 1% level. The
results suggest that the Chinese industries indeed benefit from the presence of foreign
firms through demonstration, labor training, forward and backward linkages, and
increased competition.
(5) The coefficients of absorptive capability (R) are positive and statistically significant at the
5% level in four of the six cases. In the other two cases, these are is positive but not
significant at the conventional levels (although being close to the significance at 10%
level). Thus domestic industry’s R&D seems to play some role in capturing the benefits
from FDI.
(6) It is instructive to note differences in effects of FDI variables on capital- and labor-
intensive industries. The FDI absorptive capability term (R) has a robustly positive
parameter for the capital-intensive sample, but insignificant for the labor-intensive
5
Aitken and Harrison (1999) found a similar result at the firm level for Venezuela.
9
industries. The results seem to be consistent with the following view: the certain level of
technological diffusions from FDI; but the requirement may not be necessary for the
labor-intensive industries because of either the small technology gap between MNCs and
this type of Chinese industries, or little variation in R&D over the labor-intensive
industries. In addition, the size of FDI direct effects (F) and spillovers (S) in general is
larger for the capital-intensive sample than the labor-intensive one, suggesting that the
capital-intensive industries are more responsive to the inflows of FDI and the presence of
MNCs.
(7) As expected, the domestic capital (K) parameter is robustly positive in all cases,
suggesting that domestic investment promotes industrial performance in both labor- and
capital-intensive industries. The parameter of labor input (L) shows a somewhat variable
pattern. For the entire sample, the parameter clearly lacks significance and even has the
wrong sign. For the labor-intensive sample, the sign is positive and the estimate is
significant at the 1% level, while the parameter turns to robustly negative for the capital-
intensive sample. This scenario seems generally consistent with those suggested by some
studies (for example, Barro and Sala-i-Martin, 1995), that reported varying effects of
Although the explanatory power of the model specified in Eq. (4) is good, one may worry
about the possibility of heteroscedasticity in the disturbance term and feedback from the
dependent variable to the independent variables. The White test (White, 1980) thus is used to
check whether our model may have a major specification error.6 The result of White test
6
The test is explained by White (1980, pp.824-825). The procedure consists of running a test regression of the
squares of OLS residuals from the original model on the squares and cross-products of the model regressors. Then
10
indicates that the values of the test statistic are too small to justify non-acceptance of the null
4. Concluding Remarks
Technology transfers perhaps are the most important benefit that could be brought by
multinational corporations to host economies, but they are not either guaranteed, automatic, or
free. How does a host country benefit in technology from the presence of multinational
Chinese data for the 1990s, this study seeks to advance the earlier research on the FDI-
technology transfer nexus in several ways. Besides working with a fairly large sample of 178
industries in the 1990s, we model two channels of direct effects and spillovers, through which
FDI may transfer technology to the local industries. We also directly test how host country’s
R&D may affect the technology transfers. Moreover, we take a closer look at the differential in
The main outcome of the work may be summarized in the following statements.
Technology transfers from FDI to local industries do not take place through direct effects, but
spillovers. It is so perhaps because majority of FDI received in China by the mid-1990s was in
advanced technology to the Chinese industries. Benefits from the presence of multinational
corporations seem mainly from backward and forward linkages, “learning by watching,”
competition effects, and induction of trained workers. How much technology may be transferred
under the null hypothesis, nR2, where n is the number of observations and R2 is from the test regression, is
distributed as a chi-square with degree of freedom equal to the number of regressors in the rest regression.
11
seems to be influenced by China’s absorptive capability. An industry with strong R&D ability
12
Table 1 Descriptive Statistics of the Variables Used in the Study
Full Sample (N = 178)
Q K L F R S
Mean 81.75 113.33 43.60 19.11 10.73 7.68
Std. Deviation 133.44 190.75 66.97 14.77 24.51 10.51
Maximum 909.76 1753.99 588.62 61.27 174.11 67.14
Minimum 0.28 0.25 0.22 0.00 0.00 0.02
Capital-Intensive Industries (N = 79)
Q K L F R M
Mean 116.80 162.72 38.55 22.72 16.09 10.40
Std. Deviation 173.98 255.07 44.31 16.59 28.12 12.22
Maximum 909.76 1753.99 224.14 61.13 130.15 49.08
Minimum 0.37 0.59 0.22 0.00 0.00 0.04
Labor-Intensive Industries (N = 99)
Q K L F R M
Mean 55.88 76.86 47.33 16.43 6.78 5.66
Std. Deviation 84.92 111.65 79.69 12.69 20.72 8.56
Maximum 587.83 729.06 588.62 55.87 174.11 67.14
Minimum 0.28 0.25 0.33 0.01 0.00 0.02
Notes: Q = value added of industrial output, K = capital input, L = labor input, F = foreign direct investment (FDI),
measured by the share in total capital stock in an industry, S = average share of foreign equity participation in a
manufacturing sector, weighted by output, R = FDI absorptive capacity of industry, measured by products of R&D
and FDI in an industry. Q, K, and R are in millions of the Chinese currency (RMB), F and S is in percentage, and L
is in thousands.
13
Table 2 Estimates of Technology Transfers by FDI
Variable Full Sample Capital-Intensive Sample Labor-Intensive Sample
Constant (C) 2.15*** 1.18*** 2.15*** 1.26*** 2.17*** 1.09***
(55.99) (12.78) (35.69) (9.49) (45.75) (8.65)
Capital Stock (K) 0.38*** 0.47*** 0.23***
(10.46) (7.70) (3.26)
Labor Force (L) -0.02 -0.23*** 0.23***
(-0.71) (-3.51) (3.69)
Direct Effects of FDI (F) -0.13*** -0.11*** -0.16*** -0.13*** -0.10*** -0.07***
(-10.58) (-11.42) (-8.12) (-8.49) (-6.88) (-7.28)
Spillovers of FDI (S) 0.93*** 0.60*** 0.91*** 0.67*** 0.94*** 0.51***
(39.46) (16.34) (23.09) (12.00) (34.13) (11.19)
FDI Absorptive Capability (R) 0.02** 0.01** 0.02** 0.03*** 0.02 0.01
(2.46) (2.08) (2.14) (2.67) (1.37) (1.20)
14
Appendix: List of Industries in the Sample by 37 Sectors
15
products Metal surface processing apparatus*
Foamed plastic & synthetic Daily used metal products Daily used electrical
leather* Other metal products equipment*
Plastic packaging materials 29 Machinery Lighting equipment
& containers* Boilers & engines Electrical equipment repair
Plastic footwear products Metal processing Other electrical machinery
Daily used plastic machinery 33 Electronics
products* General equipment Communications
Plastic spare parts* Bearing & valve equipment*
Other plastic products* Others in general spare Radar equipment
25 Non-metal minerals parts Radio & TV equipment*
Cement products* Forging products* Electronic computers*
Cement & asbestos General industrial Electronic apparatus*
products machinery & Electronic components*
Brick & light building equipment Daily used electronic
materials Other ordinary machinery apparatus & tools*
Glass & glass products* 30 Special equipments Electronic equipment
Pottery products Special equipment for repair*
Fire resistance products refining & mining Other electronic
Gypsum products Special equipment for equipment*
Mineral fiber products petroleum 34 Instruments
Other products excluding Special equipment for General apparatus & meter
mineral non-metallic textiles equipment
products Equipment for agriculture, Special apparatus & meter
26 Iron refining forestry & fishing equipment
Refining iron Medical equipment* Electronic measurement
Refining steel* Other special equipment equipment
Steel processing* Special equipment Calculators
Refining Iron alloy machinery repair Office instrument
27 Metal refining 31 Transportation machinery*
Heavy metal refining* Equipment for railway Watches & clocks
Light metal refining* transporting* Instruments for testing of
Precious metal refining Automobiles* electricity& electrical
Rare metal refining* Motorcycle* signals
Ferrous metal alloy Bicycles* Other apparatus & meters
Ferrous metal processing* Shipping 35 Electricity
28 Metal products Aerospace* Generation of electricity*
Metal structure Transport equipment repair Electricity supply*
Iron casing tubes Other transport equipment 36 Gas
Metal tools 32 Electrical machinery Gas production*
Metal containers & pack- Electrical machinery Gas delivery*
aging materials* Equipment for controlling 37 Water
Metal wires & transmitting Water supply*
Metal products for electricity Water delivery*
construction Electrical industrial
Note: An asterisk indicates a capital-intensive industry.
16
References
Aitken, Brian J. and A. Harrison (1999), “Do domestic firms benefit from direct foreign
investment? Evidence from Venezuela,” American Economic Review, 89 (3): 608-618.
Barro, Robert and Xavier Sala-i-Martin (1995), Economic Growth, Cambridge, MA: MIT
Press.
Blomstrom, Magnus and F. Sjoholm (1999), “Technology transfer and spillovers: Does local
participation with multinationals matter?” European Economic Review, 43: 915-923.
Borenszten, E., J. De Gregorio, and J-W. Lee (1998), “How does foreign direct investment affect
economic growth?” Journal of International Economics, 45: 115-135.
Caves, Richard (1996), Multinational Enterprises and Economic Analysis, the 2nd edition,
Cambridge, MA: Cambridge University Press.
Markusen, James and Anthony Venables (1999), “Foreign direct investment as a catalyst for
industrial development”, European Economic Review, 43 (2): 335-356.
Potterie, Bruno and F. Lichtenberg (2001), “Does foreign direct investment transfer technology
across borders?” Review of Economics and Statistics, 83 (3): 490-497.
State Statistic Bureau (SSB) (1997), The Data of the 3rd National Industrial Census of the
People’s Republic of China in 1995, Beijing: China Statistic Press.
State Statistical Bureau (SSB) (2006), China Statistical Yearbook 2006, Beijing: China
Statistical Press.
Teece, David J. (1977), “Technology transfer by multinational firms: The resource cost of
transferring technological know-how,” Economic Journal, 87 (3):242-61.
United Nations Conference on Trade and Development (UNCTAD) (2001 and 2005), World
Investment Report 2001 and 2005, New York: United Nations.
Zhang, Kevin H. (2001a), “What explains the boom of foreign direct investment in China,”
Economia Internazionale/International Economics, 54 (2): 251-274.
17
Zhang, Kevin H. (2001b), “How does FDI affect economic growth in China?” Economics of
Transition, 9 (3): 679-693.
Zhang, Kevin H. (2004), “Maximizing benefits from FDI and minimizing its costs: What do we
learn from China?” in Kehal (ed.), Foreign Investment in Developing Countries,
Palgrave/Macmillan: 78-91.
18